Improving NAS capacity and performance

My last two articles discussed the technologies being used in conjunction with traditional network-attached storage appliances to improve performance and scalability of NAS. In this final article of the three-part series, we will investigate NAS solutions that focus on addressing the scalability issues of both capacity and performance that are associated with “traditional” NAS environments. Specifically, we will look at what is called NAS aggregation or clustering.

The one remaining question with NAS systems has been how to scale performance, or throughput. As discussed in the first of the three articles (https://www.nwfusion.com/newsletters/stor/2003/0414stor1.html) TCP/IP offload engines can be used to increase the performance associated with processing the file data through the TCP/IP protocol. However, if an enterprise decides to place large amounts of capacity behind the NAS appliance, perhaps by using a storage-area network, the business applications utilizing the NAS filer are constrained to the number of interfaces and the performance of the one NAS appliance. 

NAS aggregation is one of the newer techniques being used by vendors to provide customers the ability to scale the capacity and the performance of the NAS appliance. Companies such as Spinnaker Networks, Scale8, and Zforce have created NAS appliances that, in many ways are similar to traditional NAS appliances except they have a specialized NAS operating system that allows the file system information to be distributed across many NAS “nodes.” These NAS nodes consist of processors and software, as well as storage.  Instead of adding capacity behind the appliance, these nodes are added to a cluster, automatically discovered and tied into the clustered operating system, and the storage is added to the pool of available capacity.

Much like a traditional server cluster, information is shared or distributed across the NAS nodes providing resilience in case one of the nodes should fail. As the nodes are added or aggregated, interconnect ports are also added. This means that the amount of data streams being processed increases with the addition of each node, thereby increasing the overall throughput of the “system” or cluster.

The capacity available on each node depends greatly on the vendor, as does the number of interconnect ports. However, this ability to aggregate both performance and capacity has been a huge benefit to a number of applications that require management of very large files, such as geophysical applications and video streaming/archiving.

There is no doubt that vendor support for NAS technology continues to grow, as does the market. In these three articles, we see that whether through innovation or borrowing technology from other markets, customer cries to provide NAS solutions that scale both in performance and capacity have been heard.